The Norwegian economist Trygve Haavelmo began his career as a student of
Ragnar Frisch at
University of Oslo. He went to the United States in 1939 as a
Fulbright scholar, where he ended up staying until 1947. Haavelmo spent
much of his sojourn at the Cowles Commission, before returning to Oslo. He remained a professor a the
University of Oslo until his retirement. Trygve Haavelmo won the Nobel
memorial prize in 1989.
It was during his stay in the United States that Haavelmo wrote his most
influential work -- a 1944 article introducing the "probability approach"
The "probability" approach argued that we should envision existing
economic data series as being "a sample selected by Nature", i.e.
randomly derived from a "hypothetical" series of distribution which
governed reality but which was unobservable. Thus, Haavelmo argued, we can test
the validity of economic theories by couching the theoretical model in
terms of statistical relationships which can then be tested. The relationship
between theory and the hypothetical underlying "reality", argued
Haavelmo, is akin to the relationship between the observed data and that
"reality". Therefore, if we can come close in relating theory with
observed data in some precisely defined statistical manner, then we are
effectively saying we have "reproduced" another "natural drawing"
from the hypothetical "reality" and thus our theoretical relationships
are in a sense "true".
Haavelmo also addressed the issues of identification problems in simultaneous
equations econometric problems (Haavelmo, 1943, 1947). It was Haavelmo who
differentiated the "structural" from the "reduced" form
equation and discussed the relationship between the original parameters and the
reduced-form estimates. It was Haavelmo that introduced the
"determinant condition" for identifiability. This set the
ground for the "Cowles
approach" approach to econometrics and the methodological debates on
empirical economics that raged during the 1940s.
If only we knew more about the determinants of
investment! But, unfortunately, our knowledge in this direction is still very
meager. One might well ask, What is wrong with the theory of investment? Or,
perhaps, What is wrong with the subject matter itself! For one thing, this
variable, -- the pivot of modern macroeconomics -- has apparently lived a
somewhat nomadic life among the various chapters of economic theory. Perhaps
it has not stayed long enough in any one place. Perhaps it has been ill-treated.
What is investment? Strictly speaking, investment is the
in capital stock during a period. Consequently, unlike capital, investment is a flow
term and not a stock term. This means that while capital is measured at a point in
time, while investment can only be measured over a period of time. If we ask "what is capital right
now?", we might get an answer
along the lines of $10 trillion. But if we ask "what is investment right
now?", this cannot be answered. The quantity of a flow always depends on
the period in consideration. Thus, we can answer "what is investment this month?"
(and might be told it is $10 million) or "what is investment this year?"
(and might be told $1 billion).
We can calculate the investment flow in a period as the
difference between the capital stock at the end of the period and the capital
stock at the beginning of the period. Thus, the investment flow at time period t
can be defined as:
It = Kt - Kt-1
where Kt is the stock of capital at the end of period
t and Kt-1 is the stock of capital at the end of period t-1 (and thus
at the beginning of period t).
How is the the theory of investment different from the theory of
capital? If all capital is circulating capital, so that it is completely
used up within a period, then no capital built up during the previous period can
be brought over into next period. In this special case, the theory of capital
and the theory of investment become one and the same thing.
With fixed capital, the story is different -- and more
complicated as there seems to be two decisions that must be addressed: the
amount of capital and the amount of investment. These are
different decisions. One is about the desired level of capital stock. The
other is about the desired rate of investment flow. The decisions
governing one will inevitably affect the other, but it is not necessarily the
case that one is reducible to the other.
There are effectively two ways of thinking about
the risk of annoying some people, we shall refer to these as the "Hayekian"
and "Keynesian" perspectives. The Hayekian perspective
conceives of investment as the adjustment to equilibrium and thus the
optimal amount of investment is effectively a decision on the optimal speed of
adjustment. A firm may decide it needs a factory (the "capital stock"
decision), but its decision on how fast to build it, how much to spend each
month building it, etc. -- effectively, the "investment" decision --
is a separate consideration.
Naturally, the capital decision influences the investment
decision: a firm which has $10 billion of capital and decides that it needs $15
billion of capital, therefore requires investment of $5 billion. But if this
adjustment can be done "instantly", then there is really no actual
investment decision to speak of. We just change the capital stock automatically.
The capital decision governs everything.
However, if for some
reason, instant adjustment is not possible,
then the investment story begins to matter. How do we distribute this $5 billion
adjustment? Do we invest in an even flow over time, e.g. $1 billion this week,
another $1 billion next week, and so on? Or do we invest in descending
increments, e.g. invest $1 billion this week, $500 million next week, $300
million the week after that, etc. and approach the $5 billion mark
asymptotically? Or should we invest in ascending increments, e.g. $10 million
this week, $100 million next week, etc.? Delivery costs, changing prices of
suppliers, fluctuating interest rates and financing costs, and other such
considerations, make some adjustment processes more desirable than others. These
different patterns of "approaching" the desired $5 billion adjustment
in capital stock and the considerations that enter into determining which
adjustment pattern to follow is what lies at the heart of the Hayekian approach
to investment theory.
The Hayekian approach is shown heuristically in Figure 1, where
we start at capital stock K0 and then, at t*, we suddenly change our
desired capital stock from K0 to K*. Figure 1 depicts four
alternative investment paths from K0 towards K*. Path I represents
"instant" adjustment type of investment (i.e. all investment happens
at once at t* and no more investment afterwards). Path I˘
represents an "even flow" adjustment path, with investment happening
at a steady rate after t* until K* is reached. Path I˘ ˘ is the asymptotic investment path
declining investment), while path I˘ ˘ ˘ depicts a gradually increasing investment
All paths, except for the first instant one, imply that "investment"
flows will be happening during the periods that follow t*. Properly speaking,
then, investment theory in the Hayekian perspective is concerned with analyzing
and comparing paths such as I˘ , I˘ ˘ and I˘ ˘ ˘ .
Figure 1 - Adjustment Paths towards K*
The "Keynesian" approach places far less
emphasis on the "adjustment" nature of investment. Instead, they have
a more "behavioral" take on the investment decision. Namely, the
Keynesian approach argues that investment is simply what capitalists
"do". Every period, workers consume and capitalists "invest"
as a matter of course. This leads Keynesians to underplay the capital stock
decision. This does not mean that Keynesians ignore the fact that
investment is defined as a change in capital stock. Rather, they believe that
the main decision is the investment decision; the capital stock just "follows"
from the investment patterns rather than being an important thing that needs to
be "optimally" decided upon beforehand. Thus, when businesses make
investment decisions, they do not have an "optimal capital stock" in
the back of their mind. They are more concerned as to what is the optimal amount
of investment for some particular period. For Keynesians, then, optimal
investment not about "optimal adjustment" but rather about "optimal
Major Works of Trygve Haavelmo
- "The Method of Supplementary Confluent Relations", 1938,
- "The Inadequacy of Testing Dynamic Theory by Comparing the
Theoretical Solutions and Observed Cycles", 1940, Econometrica
- "Statistical Testing of Business Cycles", 1943,
- "The Statistical Implications of a System of Simultaneous Equations",
- "The Probability Approach in Econometrics", 1944,
Effects of a Balanced Budget", 1945, Econometrica ("Supp.
- "Family Expenditures and the Marginal Propensity to Consume",
- "Methods of
Measuring the Marginal Propensity to Consume", 1947, JASA
- "A Note on the Theory of Investment", 1950,
- "The Concepts of Modern Theories of Inflation", 1951,
- A Study in the Theory of Economic Evolution, 1954.
- "The Role of the Econometrician in the Advancement of Economic Theory",
- A Study in the Theory of Investment, 1960.
- "Business Cycles II: Mathematical
- Variation on a Theme by Gossen, 1972
- "What Can Static Equilibrium models Tell
Us?", 1974, Econ
- "Econometrics and the Welfare State", 1990,
Les Prix Nobel de